AI infrastructure is about to consume vastly more resources than most people realize, and the environmental damage extends far beyond carbon emissions. A new report from the UN University Institute for Water, Environment and Health (UNU-INWEH) warns that datacenters will nearly triple their electricity consumption by 2030, while water demand will skyrocket to levels that could meet the basic needs of 1.3 billion people.

How Much Energy Will AI Actually Consume by 2030?

The numbers are staggering. Datacenters consumed around 448 terawatt-hours (TWh) of electricity in 2025, but researchers predict that by 2030, they could consume 945 TWh annually. To put that in perspective, this nearly equals the combined electricity consumption of Pakistan, Bangladesh, and Nigeria. Meanwhile, water demand is projected to reach 9.3 trillion liters per year, a volume that could supply the basic water needs of 1.3 billion people.

The challenge isn't just the raw numbers. As AI systems become more efficient, they often get used more frequently, creating what economists call Jevons Paradox. This means that efficiency gains don't necessarily reduce overall consumption; they can actually increase it. "More efficient and affordable AI and energy mean more consumption of AI, making the overall footprint far bigger than what we save through efficiency gains," explained Professor Kaveh Madani, a co-author of the UN report.

Professor Kaveh Madani, a co-author of the UN

Why Carbon Emissions Alone Miss the Real Environmental Picture?

The research highlights a critical blind spot in how we measure AI's environmental impact. Many companies and policymakers focus exclusively on carbon emissions, assuming that switching to renewable energy solves the problem. But this approach overlooks equally serious threats to water and land resources. "If we keep judging AI sustainability by carbon alone, we might think that renewables make AI infrastructure clean but that is solving one problem while creating other problems, often in places that didn't ask for it," said Dr. Miriam Aczel, the lead author of the UN study.

Dr. Miriam Aczel, the lead author of the UN study

The geographic dimension makes this even more troubling. Datacenters are frequently built near communities experiencing severe water stress, yet the people living in those areas rarely benefit from the AI systems running there. This creates an inequitable situation where local populations bear the environmental burden while distant users enjoy the technology's benefits.

Steps to Address AI's Broader Environmental Footprint

• Comprehensive Metrics: Adopt standardized reporting that measures electricity, carbon, water, and land footprints together rather than focusing on carbon alone, ensuring a complete picture of environmental impact.
• Community Involvement: Include local communities in early decision-making processes for new AI infrastructure projects, giving residents a voice in developments that affect their water and land resources.
• Investor Accountability: Treat electricity, carbon, water, and land footprints as material financial risks in AI infrastructure portfolios, encouraging investors to demand sustainability standards from companies.

Beyond energy concerns, AI also offers genuine environmental benefits that shouldn't be overlooked. In environmental chemistry, AI systems can identify water and air contaminants by analyzing complex chemical data, detecting toxic compounds and their ecosystem interactions far faster than traditional methods. AI is also accelerating the development of sustainable materials, including biodegradable plastics and more efficient industrial catalysts. What once required years of laboratory experimentation can now be completed in months.

However, these benefits come with hidden costs. The manufacturing of electronic devices and batteries used in AI infrastructure requires mining lithium, cobalt, and rare earth elements. If not managed carefully, this extraction can pollute water and soil in mining regions. Additionally, rapid technological evolution creates electronic waste as equipment is constantly replaced, adding another layer of environmental concern.

The path forward requires balancing innovation with responsibility. "The key is not to halt technological progress, but to use it responsibly," according to María M. Santiago-Reyes, a past president of the Puerto Rico College of Chemists. The combination of artificial intelligence, green chemistry, and renewable energy could help minimize negative impacts while creating more sustainable solutions.

The UN researchers recommend that all new AI infrastructure planning must consider energy, water, and land use holistically. Transparency in reporting and standardized metrics are essential for understanding the true environmental cost of AI expansion. Without these changes, the technology that promises to solve environmental challenges may inadvertently create new ones in vulnerable communities around the world.